DESCRIPTION: The hypothesis has been put forth recently that the ability of a widely used drug to eradicate a specific disease cell population can be attributed in part to the ability of that drug to form a special type of DNA adduct. The adduct attracted proteins selectively expressed in the diseased cell. It was reasoned that adducts engaged in tight complexes with proteins would be difficult to repair. Therefore the concealed adducts would persist in and selectively kill the diseased cells. The adduct shielding paradigm suggested a strategy by which a variety of aberrantly expressed proteins might be employed to increase the specificity of genotoxic compounds. Preliminary studies have demonstrated the potential of the repair shielding model as a strategy to design selective genotoxins. Bifunctional molecules were designed in which a DNA damaging warhead was tethered to a molecule recognition domain for the estrogen receptor (ER), a nuclear protein that is over-expressed in many female breast and urogenital tumors. These novel compounds formed DNA adducts that strongly attracted the ER in vitro and selectively killed cells that over-express the ER. The goal of the proposed work is to elucidate the mechanism by which selective killing occurs. The working hypothesis is that the ER binds to the toxin-DNA adducts and precludes access to the adducts by DNA repair enzymes. In the proposed studies, the removal of DNA lesions engaged in complexes with the ER will be assessed in several in vitro and in vivo DNA repair systems; these studies will determine whether the ER can indeed hinder repair of the DNA lesions formed by the novel genotoxin. Synthetic methods will be applied to vary the architectural characteristics and physical properties of ER-DNA adduct complexes to identify the molecular features responsible for selective toxicity. Incorporation of molecular recognition domains with a rang of affinities for the ER will establish whether DNA adducts engaged in tighter adduct- ER complexes are more difficult to repair (and hence more toxic) than those in weaker complexes. Variation of the DNA interactive warhead will be used to direct damage to different helical surfaces. These studies will establish the optimal molecular arrangements for adduct-ER interactions, repair shielding, and selective toxicity. In the future, combinational chemical methods could be applied to search for novel molecular recognition domains that, in principle, would permit any of a large number of disease specific proteins (including mutant forms of p53) to be used as the molecular recognition element in a second generation of toxins. These toxins could be programmed to kill cancer cells, virally infected cells, or microorganisms.